DE3718521A1 - METHOD FOR RECORDING QUASI HALFTONE IMAGES AND DEVICE FOR CONVERTING IMAGE POINT TONE VALUE DATA - Google Patents

Description

The invention relates to a method for recording of quasi-halftone images, in which a template on a Recording medium by individual recording elements reproduced with the color value "full" or "zero" will, the template in several out of individual Existing scanning fields is divided into scanning elements and each scanning field successively into the smallest subfields is divided with a predetermined size.

With such a method, templates can be created a recording medium such as e.g. B. a sheet of paper, a printing plate or a screen by individual equally large recording points, either over the entire surface colored or not colored all over are reproduced. First, the Image information of the template by a sensor or  Scanning device, such as. B. a scanner. The scanner enters for each pixel (sampling point) Signal that corresponds to the tone value (for colored originals the Hue) of the sampling point. At a Black and white template will be the shades of gray by Black to white range on the recording medium realized that the frequency of black colored recording points depending on the gray tone to be realized is varied. All Recording points of a black area of the Original are black during playback Proportion of between the black recording points arranged white recording dots in areas increases with less darkness, d. that is, the Frequency of blackened recording points is less. In white areas of the template are finally knows all the recording points, i. H. not colored. In this way, the gray scale can be in numerous discrete levels, d. H. quasi-continuous, play. In the context of such reproduced images are therefore also spoken of "quasi halftone images" with which the tonal values of the Tone scale according to a previously made division are reproduced in tonal gradations.

The processing of such quasi halftone images with the help digital technology is very simple since the image processing can be binary ("1" or "L" for one to be colored and "0" for one not to be colored Recording point). The quasi halftone corresponds to a binary recording of the image information a template that is therefore simply on a storage medium such as B. a magnetic tape, one Magnetic disk or floppy disk, but recorded  also written, for example, in a semiconductor memory can be.

In the context of the invention, the word "record" any possibility of recording the template as Quasi halftone meant. This includes in particular the reproduction of the template on a sheet of paper or on one screen, creating one for that Play the template appropriate printing plate and the Storage of the template in a memory.

One method for binary recording of templates is known from US-PS 45 78 713. With this procedure one sample scanned line by line, the tonal value of each Sampling point determined and saved. The template is divided into several scanning fields, each several Sample points include, divided, i. H. the tone value data for the sample points of the template are in data blocks divided. For each data block, the associated tone value data by averaging the mean Tonal value (average tonal value of the data block) calculated. Thereupon the number of those Determines recording points that are used to play the medium tone is required. The distribution of the recording points to be colored per scanning field by gradually dividing the scanning field into smallest sub-blocks determined. For those in every subdivision step emerging smaller blocks of data from the tonal values of the associated sampling points of the mean tone value and the number of to realize of this mean tone value to be colored Record points determined. In addition to determining the Number and distribution of the recording points to be colored an error calculation must also be carried out in each stage  be carried out with increasing Subdivision is becoming more and more complex because of the number of Subdata blocks from subdivision to subdivision increases, namely square. Because of of these increasingly extensive error calculations the advantages of digital image processing are lost.

The invention has for its object a method for binary recording of templates of the aforementioned Way of creating that without considering errors gets along.

To achieve this object, the invention provides that for every smallest subfield as well as for each of the higher-level subfields to which the relevant smallest subfield belongs from the tonal values of the scanning elements located in the subfield under consideration the mean is determined or estimated and that at least one of those considered Sub-field assigned recording elements printed if the mean is greater than a given one Reference tone value.

With the inventive method, the distribution the recording points to be colored or not to be colored, with which a scanning field of the template is shown in the playback. The Number of colors to be colored per subfield or per scanning field Recording points depend on whether the mean tone value in a scanning field or Subfield lying sampling points greater than, less than or is equal to a predetermined reference tone value. If the template is to be reproduced with the same brightness, then so  the reference tone is half the largest possible Tonal value, that would be the case with black and white originals Reference tone value is a tone value that is exactly on the Half between the tonal values for white and for black lies. Should the original be reproduced darker (lighter) the reference value is smaller (larger) than the mean of the tone scale by reproduction. By Raise the largest tonal values in a template and / or by lowering the smallest in the template Occurring tonal values can amplify the Contrast of the original achieved in the recording will.

In the method according to the invention, first of all the smallest subfields in which a scanning field of the Submission is divided. That means that initially considered the tonal value data of the smallest data blocks will. For every smallest block, the corresponding one Tone value data the mean tone value, i.e. the Average determined. If this mean is greater than the reference tone value, at least one of the considered recording points assigned to the smallest subfield printed, or more precisely, as to printing record point considered. Thereupon are the next largest subfields, each from a certain number of smallest subfields put together, considered. From the tone value data of the corresponding to these next larger subfields The average tone value is determined for data blocks. Is the mean value is greater than the reference tone value, so one of the next larger subfields under consideration assigned recording points printed. In addition to the in Dependence on the mean tone value of the scanning elements of the smallest subfields of printed recording points  is going to be the next largest Subfields another record point printed when the mean tone value of the next largest Subfield is larger than the reference tone value. In the The next phase of the process will be the next larger subfields in the previously described Way examined until finally in a last Phase of the procedure examined the scanning field itself becomes. Is the average of all in a scan field lying scanning elements larger than the reference tone value, becomes from the set corresponding to the scan field Recording points one printed.

On the one hand, there are so many recording points per scanning field printed that the tonal value of each On average, the scanning field of the original is reproduced exactly becomes. On the other hand, the recording dots to be printed distributed over the scanning field in such a way that the original with the least possible loss of sharpness or in other words with the most accurate reproduction of details is recorded. For good reproduction a template is besides the fact that the tonal values of the original are almost exact reproduced, also crucial, that the template in detail, d. H. reproduced with maximum resolution becomes. The division of the template into the Scanning fields leave only a relatively low resolution the template to; the division of the template into the However, scanning fields only has the task of Minimum requirements regarding the mean tonal values sure to meet d. H. at least the tonal values related to the scanning fields of the original almost exactly play. The desired resolution and so that a reproduction of the template with the desired  Sharpness is achieved by starting from the smallest subfields for each subfield on the Basis of the mean tonal value of the subfield from the Amount of recording points assigned to the subfield at least one is printed. Why the detail reproduction of a template with increasing Subdivision of the scanning fields into smaller and smaller fields increases the accuracy of detail reproduction, is explained below using an example will.

Experience has shown that the human eye can do up to 150 tone levels (i.e. in the case of a black and white 150 shades of gray from black to white) visually differentiate. The tonal value of a scanning field should reproduced by one of 150 tone levels with the scan field (i.e., the areas the template within which the human eye tonal values can not be selected too large may. Since the template only by individual recording points reproduced with the color value "full" or "zero" a scanning field must pass through at least 150 Recording points are reproduced. Depending on the Number of recording points colored for this scan field there are 150 tone levels.

The digital technology common today for image processing therefore suggests that for the size of the scanning field this comprises 256 sampling points or 256 recording points is played. The sampling or Recording points are each in one (16 × 16) matrix arranged. The mean tone value of a such a scanning field can be recorded almost exactly through one of 256 levels of tone (i.e.  with an accuracy of 1/256). For example, about 20 µm as the diameter of a scanning point based, you get a scanning field of size 300 µm × 300 µm (16 × 20 µm = 320 µm ≈ 300 µm). An inch the template could be roughly thirty next to each other or scanning fields arranged one below the other be divided. With the division of the template into the scanning fields can only have a resolution of approximately 30 lines / cm (30L / cm) will. This resolving power is of course too low to the requirements in terms of sharpness and the detailed reproduction of the template. Around to be able to achieve greater resolution, the scanning field is divided into several smaller fields divided. For the subfields of the scanning field takes the resolving power depending on the size of the subfields, in which the scan field is divided to. For example enables the scanning field to be divided into four Subfields twice the resolution of the scanning field, a further subdivision of the individual subfields in four subfields of the same size four times the resolving power, etc. In the borderline case the scanning field can be divided 256 times, i. H. by 256 recording points are reproduced, the highest possible resolution, in the exemplary Fall 500L / cm is reached. In this case, the Template can be reproduced in 20 µm wide lines. A 1 cm wide strip of the template would go through 500 lines with a width of 20 µm on the recording medium are reproduced.

In the method according to the invention, no need Error considerations or calculations carried out will. Based on the individual sampling points or  the individual smallest subfields are the pro Scanning field (based on the example above) 256 scanning-side signals with 256 signal levels each, which are required to receive each sample signal (each sampling point) a tone value between 0 and 255 assign, transformed into 256 binary recording signals, that are either "High" or "Low". The means that to save a template as a quasi-halftone image per scanning field instead of that for the scanning side Signals required 256 bytes (256 × 8 bits) only 256 bits are required. The necessary Storage requirements are reduced by the Factor 8. Determining whether the mean tone value is off the scanning-side signals belonging to a subfield larger than the reference tone value, for example larger than the mean tonal value that can be represented, can be based on the scanning-side available in digital form Signals are carried out very quickly. Based on the most significant bits of the tone value sum from the binary-coded scan-side signals of a subfield can determine whether the mean tone value for the Subfield larger than the average tonal value that can be represented is. If the most significant bit is a "1", it means that the mean is greater than the largest with the The number of binary numbers that can be represented divided by 2, d. H. larger than the middle representable Number. Averaging is therefore not necessary to be completed; it is enough if only considers the most significant bit of the sum and the Average value is estimated on the basis of this bit. Under "Averaging" is also an "averaging" to understand.  

A major advantage of the method according to the invention is its simplicity, which is accordingly simple and allows fast working embodiments. Its application is therefore first of all if a picture template is "online" and fast, e.g. B. in a scanner with about 1 MByte / s maximum data transfer, or with a very high, half-tone resolution, for example, 500 L / cm shall be. It also offers itself but also when using the digital ones available today Transmission channels, e.g. B. digital copier or Fax, not just line, graphic and text templates, but also transfer raster and halftone images should be. After all, the application is also for large digital displays, e.g. B. LCD displays which have the same requirements, appropriate. The method can also be used for this grid and line templates with low distortion to be transferred to binary record carriers especially if a high-resolution image is used to capture the template Scanning device is used and the binary recording points are sufficiently small.

The location is advantageously dependent on from the amounts of the mean values of the recording elements to be printed same within the subfields Order determined in a random selection, the Location of depending on the amounts of the averages recording elements to be printed in the subfields different order is determined such that each recording element is printed once. The Location of the recording points belonging to the respective subfields, which depending on the amount the average tonal value of each subfield are printed,  is chosen at random within equally sized subfields. It is crucial that within higher-ranking subfields, each consisting of lower-ranking smaller ones Composing subfields, the location of the potentially too printing recording points is set so that each record point is printed at most once becomes. The location pattern for those in different sizes Subfields of record points potentially to be printed would be a full color when overlaid Field result. By dividing the depending on the amount of the mean tonal values of the individual subfields of the same and different Size ensures that each tone value of the original is reproduced almost exactly.

An advantageous embodiment of the invention exists in that to record the template with a Tone gradation that is coarser than that of the original which is dependent on the mean tonal values of the scanning elements to be printed only be printed for such subfields, their number on scanning or recording elements larger or same as that for displaying the template with the desired one Tone gradation is the required number. If the tonal gradation in the playback is not so needs to be fine, like that of the template, then need the subfields only from a certain one Minimum size to be examined whether in Dependence on the average tonal value of the subfield a record point is to be printed. For example, should instead of the 256-step tone scale reproduced a 16-step tone scale are used to determine the number and the Location of recording points to be printed only partial fields  (Data blocks) that are at least as large as 16 Are sampling points (the minimum 16 tone value data). This increases the Processing speed of the quasi-halftone image.

This measure makes the method particularly easy just do it. With regard to image processing digital technology is preferred it is particularly useful to divide the scanning field by means of digital-value steps, d. H. the tone value data for the scanning points of the scanning field first in two equally large first blocks, these the first two blocks then into two of the same size Blocks (there are then four data blocks of the same size before) etc. to disassemble. The division into each the same number of data blocks of the same size also have an effect beneficial to data management and mapping the data for the individual data blocks.

The size advantageously corresponds to the smallest Subfields of the template, d. that is, the number of one smallest data block summarized tonal values the Ratio of recording to sampling fineness or -resolution. The smallest subfields have the size two recording points when the recording fineness greater than or equal to the scanning fineness is or is selected. A ratio of "1" states that a sample point of the original is also a Record point in the playback corresponds. From that a distinction is made between the enlargement ratio between submission and reproduction, which is not considered here becomes. So a ratio of "2" would mean that a sampling point by four recording points  is pictured. For those subfields that are smaller or the same size as a sampling point, becomes the tonal value of the sampling point in which the subfield is taken as the mean. So here one is omitted separate mean calculation. With a relationship "2" would be the mean of two neighboring ones smallest subfields equal to the tonal value of the Sampling point, in which the two smallest subfields lie.

It can be beneficial to record more precisely than to feel. The small scanning fineness has that Advantage that with sufficient sharpness less Tone value data accrued. The higher recording fineness has the advantage that the recorded image has a has little grain, which makes it appear calmer and can be viewed better. Overall fewer addition steps per matrix are required since the mean tonal values for those subfields that are smaller than or the same size as a sampling point, no averaging (i.e. sum formation) required is. If the recording is finer than Sampling can already be based on the most significant bit the digitized tone value of each sampling point, the is represented by two recording points and is equal to the size of the smallest subfields whether one of the two recording points is too print is. In this case it would be about 50% less Adding steps per scanning field may be required. At recording twice as fine as sampling could even be 75% of the adding steps per scan field can be saved because of the most significant bits of the digitized tone value of a sampling point can be whether a recording point in  the smallest subfield is to be printed while using of the second most significant bit of the digitized tone value can be determined whether a recording point is to be printed in a next larger subfield.

The sampling and recording points are advantageous each in an orthogonal matrix arranged. With such a Arrangement of the sampling points very simple in individual Subfields are divided, with the individual sampling points very easily assigned to the individual subfields can be. This is also an easy one Assignment of the individual tone value data to the different and different sized data blocks given, which simplifies data management.

Preferably, the sample and record points are each arranged hexagonally. With this Ordering scheme of the sampling or recording points has a sampling point six so-called closest neighboring points on, d. H. the sampling or recording points of a series are each "on Gap "between the sampling or recording points of the neighboring rows. The lines or Columns of such a matrix are therefore alternating shifted against each other. By such a thing Ordering scheme can on the one hand be the template scanned much more precisely and on the other hand the image are recorded much more precisely because of the size the gaps between the individual sampling or Recording points due to the "high packing density" of the sampling and recording points, each one Show circular area is minimized. This hexagonal  Ordering scheme is especially with small scanning and Record points advantageous.

The method according to the invention cannot only be used for recording of single color, but also for recording of multi-colored quasi-halftones that are made up of put together several additive or subtractive primary colors, be used. This will be the above described procedures performed for each of the primary colors and the resulting single-color quasi-halftone images then overlaid. For every additive or subtractive basic color is at each sampling point Sampling a certain tone value assigned to the basic color. Subsequent subdivision of a scanning field is the mean tonal value of the respective Basic color of the created subfield in the above described way determined or calculated. The individual recording points can either be full with the corresponding additive or subtractive Base color can be colored or not colored at all. With multi-color image recording with the help of this Procedures can also avoid the known techniques of Moir appearances.

The method can also be used to record three-dimensional Templates are used. Doing so an original on a recording medium individual recording volume elements with the color value "Full" or "zero" three-dimensionally represented that Template in several from individual sample volume elements existing sample volumes divided and each sample volume successively into the smallest partial volumes with one predetermined volume divided.  

When recording three-dimensional originals as Quasi halftone images are provided according to the invention in that for every smallest partial volume as well as for each of the parent sub-volumes to which the relevant heard the smallest partial volume from the tonal values of the in the sample volume elements lying under consideration the mean is determined or estimated and that at least one of the recording volume elements, the partial volume just considered are assigned, reproduced with the color value "full" if the mean is greater than a given one Reference value.

The further developments or variants described above of the method for recording two-dimensional Templates also apply analogously to the case of recording of holographic images. The related considerations made with the further training just have to on the three-dimensional Space to be transferred.

To record holographic images, the Scanning and recording volume elements each either into an orthogonal three-dimensional matrix or in a three-dimensional matrix with hexagonal densest ball packing can be arranged. While the The first option is an easily understandable subdivision the three-dimensional template in the first, second, third, etc. volume allowed, offers the second possibility the advantage that the three-dimensional Submission by the arrangement of the scanning or recording volume elements with less loss of information can be recorded or played back; because with the hexagonally closest spherical packing it is  Size of the spaces between the individual volume elements minimal.

The invention also relates to a device for converting pixel tonal values, with a processing unit, which tone values one into several Sample fields divided into recording pixel data converted to either the color value Correspond to "full" or "zero".

According to the invention, this device is characterized by
an input switching network which receives the pixel tone value data and combines them into a subdivision of the scanning fields of the template into partial fields according to pixel tone value data blocks,
an arithmetic processor connected to the input switching network, which determines or estimates the average tonal values from the pixel tonal value data of each block and outputs an output signal for each block if the mean tonal value of the block is greater than a predetermined reference value, and
an output switching network connected to the output of the arithmetic processor, which, depending on the output signals of the arithmetic processor, creates the recording pixel data or the distributions of the recording pixel data corresponding to the color value "full" and the color value "zero".

On the input lines of the input switching network all pixel dot data stand for Scanning field. The order in which the pixel Tone value data on the input lines is in  swapped the input switching network so that the Tone value data combined into smallest data blocks, the smallest data blocks to the superordinate data blocks summarized etc. on the output lines of the input switching network. By doing Input switching network is therefore regrouped the tone value data in the individual data blocks. This regrouping corresponds to the division of the scanning field in the smallest subfields and in higher-level subfields of different sizes. The exits of the input switching network are with the inputs of an arithmetic processor connected from the mean tone values for the individual tone value data the data blocks concerned are determined or estimated. The arithmetic processor inputs for each data block binary output signal that, for example, "high" is when the average tone value of a data block is larger as a predetermined reference value, and is "low", if the average tone value of a data block is smaller or is equal to the specified reference value. The Arithmetic processor thus receives the tonal value of each Sampling point of a scanning field in the form of a binary coded Signals and returns for each data block, i.e. H. for every smallest data block, for everybody from the smallest data blocks composing next larger Data block, etc. a binary output signal. The Output signals from the arithmetic processor are called Input signals in the input switching network given. The output switching network determines based on of its input signals, which recording point per Subfield is printed when the output signal of the Arithmetic processor for the one corresponding to the subfield Data block is "high". The output switching network gives for each recording point of the  Scan field a binary signal that either "High" (for example for a dot to be printed) or "Low" (for a point not to be printed). The pixel tone value data according to the invention Processing device transforms as the binary coded Tonal values of the individual sampling points Scanning field into binary signals, either the Can assume state "high" or "low" and indicate whether or not to color a recording point. All processing is digital and can can be executed in parallel and therefore very quickly.

In an advantageous development of the invention the arithmetic processor has a multi-level network from interconnected adders with carry output has, being in the lowest level as many adders as the smallest data blocks and each lowest level adder is the Tone value data of the smallest data block belonging pixel added. The outputs of the adders one level with the inputs of the adders connected to the next higher level while the carry outputs the adder of each stage outputs the Arithmetic processor are. Here is the Arithmetic processor as a multi-level network Adders with carry output implemented. The carry outputs the adder of a stage represent the outputs of the arithmetic processor of the corresponding data blocks So are the carry outputs of the adders the lowest level, each of which is the tone value data of the add the smallest data blocks, the outputs for the smallest data blocks. The entire arithmetic processor is very clear and consists of  commercially available elements, which is why it is very inexpensive can be manufactured.

Advantageously, the input and the Output switch from a programmable Control unit controlled. Here is the input switching network controlled such that the pixel Tone value data from the inputs of the input switching network according to the division into data blocks to the outputs of the input switching network are switched. In addition, the control unit controls also switching the input signals to the Output switching network on its output lines.

In a preferred embodiment of the invention the type of aggregation of the pixel tone data to data blocks and the type of distribution of the recording pixels in one with the control unit connected memory, being stored in the memory several different summarization and distribution patterns are saved. So is in the memory at least one control pattern for each Input and for the output switching network saved. Are several such control patterns stored for both networks, the Division of the tone value data of a scanning field into Data blocks and on the other hand the distribution of the pro Block of data to be printed from Scanning field to scanning field can be varied.

In an advantageous development of the invention it is provided that the pixel tone value data of a scanning field, which comprises pixels arranged in a matrix with 2 ⁿ rows and 2 ⁿ columns, are present at the inputs of the input switching network, where n is a natural number and the tone values the pixels are present as a binary number with 2 n digits, and that the adders of the arithmetic processor are digital adders for adding 2 ⁿ -digit binary numbers. If a digital value number of sampling points is combined to form a sampling field, that is to say that a digital value number of pixel tone value data is combined to form a (main) data block and fed to the input switching network, the arithmetic processor provides a particularly simple estimate of the Amount of the mean value of the tone value data belonging to the respective smaller data blocks. This is because the mean value can be estimated on the basis of the most significant bits at the output of the digital adders, ie on the basis of the carry bits of the digital adders. A real mean calculation is not necessary. This saves computing time.

The number of carry outputs of each adder corresponds the number of pixel dot data per smallest data block or the number of one subordinate data block belonging subordinate Data blocks.

The arithmetic processor is advantageously such controllable that if the recording fineness is greater than the scanning fineness, the input lines of the Arithmetic processor for the pixel tone data connectable to the inputs of the adders of higher levels are, the input lines of the Arithmetic processor for the most significant bits of the Pixel tone data with the outputs of the  Arithmetic processor are connectable. Is the fineness of recording greater than the scanning fineness the mean value calculation for the smallest or smaller blocks of data by examining the most significant bits of the pixel tone data of the smallest or smaller data blocks are replaced. The most significant bits are then sent directly to the outputs of the arithmetic processor for those concerned Data blocks occupied.

When the recording fineness is greater than Sampling fineness are the input lines of the Arithmetic processor for the pixel tone data with the inputs of the second stage adders and additionally the input line of the arithmetic Processor for the most significant bit of the Pixel tone data with the outputs of the Arithmetic processor connected. At a 2 times recording fineness greater than scanning fineness the input lines of the arithmetic processor for the Pixel tone value data with the inputs of the adders the third stage and the input lines of the Arithmetic processor for the most significant Bits with the outputs of the arithmetic processor for the smallest data blocks and the input lines of the Arithmetic processor for the second most significant Bits with the outputs of the arithmetic Processor for the parent of the smallest data blocks Data blocks connected.

The smallest data blocks in which the input signals of the input switching network are summarized, each contain two tone data if the Fineness of recording greater than or equal to  is the scanning fineness. The input switching network is controlled so that each input of the Input switching network for a tone value data signal with a the ratio of recording to Fineness corresponding number of outputs of the Input switching network is connected. With variation the ratio of recording to sampling fineness so does not need, as in the above variant, the Arithmetic processor to be switched; the Rather, switching takes place in the input switching network, in such a way that at the entrances of the arithmetic processor to be processed per data block Tone data signals grouped correctly issue.

With a ratio of is an input of the Input switching network on which a tone data signal is present, with two outputs of the input switching network connected. These two output signals set the tone value data for a smallest Data block. With a ratio of "2" is a Input of the input switching network at which a Tone data signal is present, with four outputs of the Input switching network connected. These four Output signals correspond to two smallest data blocks with two tone value data each. By appropriate Controls of the input switching network can binary coded tone data of the sampling points of the Submission at any ratio from record to Scanning fineness in the associated binary signals, which can only assume the "high" or "low" state, be transformed.  

The following are with reference to the figures Embodiments of the invention explained in more detail. in the individual show

Fig. 1a to 1i an example of the division of a scanning field into individual partial fields, that the subdivision of the tone values of the individual scanning points of a scanning field containing the (main) data block in the individual sub-blocks of data,

FIGS. 2a to 2i those recording dots to be printed in response to the magnitude of the mean value of the tone values of the individual subfields,

Fig. 3a to 3f is an example for recording a two-dimensional template if the scan definition is equal to the recording fineness,

FIGS. 4a to 4f Examples for recording a two-dimensional template, if the recording definition is times or 2 times larger than the scan definition,

Fig. 5 is a block diagram of the apparatus for converting pixel tone values in binary data signals,

Fig. 6 shows the structure of the arithmetic processor of FIG. 5,

Fig. 7a to 7e is an example of the allocation and distribution of the comprehensive to be printed recording dots in a 4 × 4 sampling points scan field,

FIG. 8 shows an example to explain the operation of the device shown in FIG. 5 and the arithmetic processor shown in FIG. 6 and

Fig. 9, the fitting in the case of the example of Fig. 7 and 8, the adders and the inputs and the outputs of the arithmetic processor digital signals.

In Figs. 1a to 1i each subdividing steps for recording a two-dimensional template are shown for the case that the sampling resolution is equal to the recording resolution. This means that a scanning point of the original is reproduced by a recording point on the recording medium. For the sake of clarity, the sampling and recording points are each represented by square fields spaced apart from one another. In fact, the sampling and recording points have a circular area. Furthermore, for the sake of simplicity, assume that the original to be recorded is a black and white image. In this case, the tonal value of a sampling point corresponds to a certain shade of gray (gray level) between the "gray values" black and white.

The original to be recorded is in a scanner, such as B. a scanner, an electronic Camera in which the scanning-side image signals (the signals for the individual sampling points) one have a good signal-to-noise ratio (e.g. 100: 1 in Voltage amplitudes) sampled line by line. That from of the scanner for each sample point Signal represents the (analog) tonal value of the concerned Sampling point. These signals are in one Digitized analog / digital converter and then stored so that they can be used throughout the procedure are always available. The others Calculations are done in a computer. The template will on the recording medium of a recording device, for example on a sheet of paper, a printing plate or a screen, by individual Points (recording points) reproduced or shown. The recording device receives the Record required signals from the computer.

The template is first shown in FIG. 1 in divided scanning areas 12 each consist of several sampling points arrayed fourteenth The coordinate axes X and Y , by means of which the position of a scanning point 14 in the matrix is indicated, are shown in FIG. 1a. The same coordinate system also applies to the matrix of the recording points 16 ( FIG. 2). When reproducing the original on the recording medium, that is to say when copying the original, care must be taken that the average tone value of a scanning field (ie the average value of the tonal values of all sampling points 14 of a scanning field 12 ) is reproduced as precisely as possible by the recording points. As already mentioned in the introduction, approximately 150 tonal value levels (grayscale levels) can be visually distinguished from the human eye, so that each scanning field 12 of the original must be represented by at least 150 equally large recording points on the recording medium in order to represent them. A single recording point can only assume the color value "full", that is black, or the color value "zero", that is white. The different grayscale levels between white and black can be realized by varying the number of colored recording points within an area comprising at least 150 recording points.

The digital technology commonly used in image processing today suggests dividing the gray scale into 256 levels. The 256 gray levels can be represented on the recording medium by 256 recording dots 16 arranged in a matrix having 16 rows and 16 columns. That is, a scanning field 12 of the original on the recording medium is represented by 256 recording dots arranged in a matrix. The mean tone value of a scanning field 12 can therefore be reproduced practically exactly with an accuracy of 1/256 = 1/2 ^-8 .

To determine the average tone value of a scanning field 12 , the analog signal, which is provided by the scanning device for each sampling point and represents the (analog) tone value of the relevant sampling point, is digitized in an 8-bit analog / digital converter with 256 signal stages. With the 8-bit binary code, the (continuous) gray scale of the original can be subdivided into 256 gray levels in the case of a black and white original. The division of the gray scale is so fine that the human eye cannot or hardly distinguish the difference between two gray values when looking at a scanning field. The continuous gray scale of the original can be reproduced quasi-continuously in the recording.

The method now serves 256 signals on the scanning side each with 256 signal levels in 256 binary recording signals to transform, with sufficient Quality in terms of tonal and Detail rendering.

In the case of the brightest tonal value (level 0) it is advisable not to print a single recording point, in the case of the darkest tonal value (255) the 255 points obtained above are all accordingly. What happens with the 256th dot (provided with the reference number 18 in FIG. 2a) is insignificant because the eye cannot see very small tonal value differences in normal tonal distance in dark tonal ranges. Therefore, the 256th point can be treated as well as the 255th, as a "colon" with it, or not used at all, ie never printed.

Obviously, only one binary recording point 16 can be assigned to each individual sampling point 14 , so that its tonal value is normally recorded too high, namely printed, or too low, namely unprinted. However, if one disregards point-by-point image modulation and instead looks at areas that consist of several pixels, the finer tonal gradations are obtained with increasing area. The desired fine tonal gradation (256 + 1 steps) is only obtained for the entire recording area consisting of 256 pixels. This is sufficient because at 20 µm dots, the total recording area (consisting of 256 pixels) is still so small (side length approx. 16 × 20 µm = 320 µm = 0.32 mm) that it is visual at normal viewing distance (30 cm) is just recognizable. A good TV monitor can serve as a comparison, the pixels of which with HDTV technology also have a tone value gradation of up to 8 bits, with a luminous dot spacing of approximately 0.30 mm. However, while the recording dots in the printed image abut each other seamlessly, the luminous dots of the monitor are separated from each other, which is no longer recognizable. The method is characterized by the fact that the limited performance of the eye is used several times to enable a relatively simple technical solution.

While FIG. 1 shows an example of a possible subdivision of the scanning field 12 into nested subfields, FIG. 2 shows which of the recording points 16 belonging to a subfield are printed when the mean tone value of the scanning points 14 in this subfield is a predetermined reference value exceeds. The recording dots to be printed are each represented by an "×". To get to the subdivision shown in Fig. 1, the scanning field 12 shown in Fig. 1a is first divided into two equal halves Fh ₁ and Fh ₂ ( Fig. 1h), the two halves each in turn into two equal halves Fg _1,1 , Fg _2.1 and Fg _1.2 , Fg _2.2 ( Fig. 1g) and so on. So there is a successive subdivision of the scanning field 12 into two halves, four quarters, eight eighths, etc. until finally after seven subdivisions 128 smallest subfields Fb _ÿ , with i = 1, 2 _,. . ., 8 and j = 1, 2 ,. . ., 16, which consist of two scanning or recording points and each make up 1/128 of the total area of the scanning field. The scanning field 12 is then divided into several subfields of different sizes and nested within one another. This means that the tonal value data of the sampling points 14 of the scanning field 12 are divided into several data blocks of different sizes and also nested within one another. The distribution of the recording points 16 to be printed as a function of the average tonal value of each subfield, as shown in FIG. 2, is only one example. There are innumerable possibilities for the location of the recording points potentially to be printed per subfield. Care must be taken, however, that a recording dot potentially to be printed in a larger subfield does not coincide with a recording dot potentially to be printed within a smaller subfield. That is, if the recording dots to be printed shown in Fig. 2 were superimposed, all but one of the recording dots 16 would be colored. The example given in FIG. 3 will be used to explain when and at what point a recording dot 16 is printed as a function of the tonal values of the sampling points 14 .

For the example it is assumed that the template should be reproduced with the same brightness. This means that a scanning field which has an average gray value of 150, for example, is also reproduced with a gray value of 150 on the recording medium. For this purpose, a total of 150 of the recording points 16 corresponding to the scanning field 12 would have to be black. If the original and the reproduction are to be of equal brightness, the reference value above which the mean value of the tonal values of the scanning points 14 of a subfield must lie in order for a recording point 16 belonging to the subfield to be printed is equal to the mean tonal value which can be represented with the 255 recording points per scanning field , ie 127.5 (note that for the 256 shades of gray from 0 to 255, only 255 of the 256 recording points are required, so that one recording point per scanning field is not required and can always be white). The reference value expediently corresponds to the gray tone 128, which will be explained later.

In Fig. 3a, the gray values are shown for the 16 sampling points in the top left in the scan 12th The method is only to be described in detail here on the basis of these 16 sampling points; for the remaining sampling points 14 of the scanning field 12 , the procedure is analogous. When assessing which of the recording points 16 corresponding to the scanning field 12 are printed, the individual scanning points or their gray values are assumed. First, the tonal values of two adjacent sampling points combined to form a smallest subfield Fb _{i, j} are considered, as is shown in FIG. 3b. For each of the smallest subfields, the mean tone value (mean value of the tonal values) resulting from the tonal values of the two sampling points of the smallest subfield is first calculated. If this mean is above the gray value 128, one of the two recording points 16 belonging to the smallest subfield in question is printed in accordance with the pattern shown in FIG. 2b. The gray values of the individual sampling points are available as an 8-digit binary code. In order to decide whether a recording point is to be printed, the binary codes of the corresponding tonal values are used. The following calculation results for the smallest subfield Fb _1.1 of FIG. 3b:

In digital circuit technology, the mean is of two numbers expediently by an addition and formed a shift step. For the case above therefore results for the mean

L000 L00L, 0

At the same time, it is clear that the carry bit of an 8-bit adder assumes the value "L" if the sum of the two numbers exceeds 255, ie their mean value is greater than or equal to 128. It is therefore sufficient to carry out an addition in order to be able to use the carry bit to indicate whether the mean value of two numbers reaches or exceeds the predetermined limit value. The application of this knowledge is initially suitable for the smallest subfields of FIGS . 1b and 3b, because the smallest subfields serve to represent the most significant bit of the mean of the two tone values.

For the other smallest subfields of FIG. 3b, the following results analogously:

Recording points are therefore only printed in the subfields Fb _1.1 and Fb _2.1 , since in both cases, when the two respective tone values are summed, a carry bit arises which has the value "L". According to the recording point scheme shown in FIG. 2b, the recording points with the X , Y coordinates 2, 1 or 3, 1 are printed or addressed.

To determine which recording points are to be printed in the 128 smallest subfields according to FIG. 1b or FIG. 3b, only 128 adding steps are thus required. A single 8-bit adder using low-power Schottky technology requires, for example, 25 nanoseconds (ns) for an adding cycle (SN 74 S 283 from Texas Instruments), which corresponds to an execution frequency of 40 MHz. In order to calculate the pressure points for the 1/128 of the subfields as quickly as possible, you can first connect several individual 8-bit adders in parallel and, for example, get 16 add cycles for 8 adders connected in parallel, i.e. 400 ns or 2.5 MHz for the above listed type. Much faster processing options are achieved with so-called customer-specific integrated circuits, the adders of which sometimes exceed execution frequencies of 100 MHz. The data transfer rates of 1 Mbyte / s required for the online application of the method in modern scanners can thus be achieved and clearly exceeded.

In the next step, the average tonal values of the subfields drawn in FIGS. 1c and 3c, which each make up 1/64 of the entire scanning field, are calculated. First, it is checked whether the mean value of the four tone values is greater than or equal to 128, ie whether the sum of the four tone values is greater than or equal to 2 × 256 = 512. However, the sum of four binary-coded numbers is already greater than or equal to 512 if the carry that arises when the seven least significant bits of the four binary numbers are added has the value "2L". The subfields Fc _{i, j of the} second order are used to represent the second most significant bit of the mean value of four tone values. It is therefore not necessary to look at the entire eight digits of the four binary-coded tonal value data, but it is sufficient to add up to and including the 2⁶-valued digit. The sum of the four tone values can be formed from the individual tone values, namely by means of three addition steps. However, it is more favorable to calculate the sum of the (already calculated) tonal value sums of those smallest subfields Fb _{i, j} shown in FIG. 3b, each of which lies in one of the superordinate subfields Fc _{i, j} shown in FIG. 3c. This is because only one adding step is required. However, the sum of these two tone value sums is greater than or equal to 512 (the mean value is greater than or equal to 256) if and when the eight low-order bits of the two tone value sums are added, a carry bit with the value "L" results. The subfields Fc _{i, j of} second order also serve to represent the most significant bit of the mean value of the two mean tone values formed from two tone values in each case. It is surprising that it is not necessary to determine whether the tone value sum of the four tone values is greater than or equal to 512, that the most significant carry bits resulting from the calculation of the tone value sums of the smallest subfields need to be taken into account. In both cases, neither the higher nor the lower-order bits of the tonal value sums calculated in the first step apparently play a role in determining whether the sum is greater than or equal to 512. Therefore, an 8-bit adder is also sufficient for the summation.

The following calculations then result for the second-order fields shown in FIG. 3c:

In the above example, a carry bit with the value "L" only arises when the tonal value sums for the field Fc _1,2 , which is composed of the two smallest subfields Fb _1,3 and Fb _1,4 , are calculated. This means that a recording point is only to be printed in the subfield Fc _1,2 . According to the recording point pattern indicated in FIG. 2c, this is the recording point with the coordinates 2, 4 which is shown in FIG. 3c. In order to determine which recording points are to be printed within the second-order subfields, only 64 addition steps are required.

The above considerations can also be applied analogously to the higher order subfields. Next, the subfields Fd _1.1 and Fd _{1.2 of} third order drawn in FIGS. 1d and 3d are to be examined. If the average tone value of the sampling points belonging to each individual third-order subfield is greater than or equal to 128, ie the sum greater than or equal to 4 × 256 = 1024, one of the recording points belonging to the relevant third-order subfield must be printed. The tone value sum is greater than or equal to 256 if and only when the eight low-order bits of the tone value sums of the associated second-order subfields are added, a carry bit with the value "L" arises. In the example, for the two subfields Fd _1.1 and Fd _1.2 third order:

Since the carry bit has the value "L" for both sub-fields of the third order, a recording point of the eight recording points belonging to a sub-field must be printed. The recording dots to be printed in accordance with the recording dot pattern shown in FIG. 2d are entered in FIG. 3d. To determine whether one of the recording points belonging to the sub-fields of the third order is to be printed, only 32 addition steps are required in the third stage.

For the subfield Fe _{1.1 of the} fourth order shown in FIG.

Since the carry bit has the value "L", one of the recording dots belonging to the subfield Fe _1,1 must be printed. The recording point resulting from the recording dot pattern according to FIG. 2e is shown in FIG. 3e. In Fig. 3f all, the upper left corner of the scan field 12 are corresponding, located to be printed recording dots.

If the method is applied to the entire scanning field 12 , the position of the recording dots corresponding to the scanning field 12 to be printed is obtained analogously to the procedure described above.

Generally speaking, the process is used to from the scanning-side (analog) pixel tone value signals binary signals that are either "high" or "low" are to generate. In this way, an "analog" Image "binary" can be recorded. The record must not necessarily in the form of a print of the picture; the fact that the picture is in the form of individual binary signals, the easy storage of the image in a magnet or Semiconductor memory. It is also a matter of saving  to record the image in the sense of the present Registration.

There are no error calculations in the method described here or considerations required. The Number of recording points per scan field are printed so that the tonal value of the template is "sample-by-sample" is reproduced exactly, need not in to be calculated in advance, but results automatically when distributing the recording points on the subfields of the different orders. If is examined for each of the 128 smallest subfields, whether to print a record point belonging to it it is not yet clear how many recording points overall still have to be printed so that the Scanning field both in terms of its mean tone value as well as in terms of detail reproduction or the - Contrast is reproduced true to the original. Nevertheless are the same number of recording points at the end of the process printed as it is for exact reproduction of the mean tone value of the scanning field is required is. Because the scanning field is nested into smaller and smaller ones Subfields is divided and the Distribution of the recording points to be printed for every single subfield of the different orders is made, the individual details of the template both with maximum sharpness and with maximum Contrast reproduced.

The reliability of the process is based on the two examples described below clearly. It has shown that the reproduction of image areas, that have a tone value of 127 or 128, especially can be faulty. The tone values are 127 and 128  critical because it goes through the middle two the 256 recordable tonal values correspond. A sample point with a tone value of 127 or 128 can only be replaced by a white or a black Recording point can be played back. The one at the Recording of a single sampling point So error is maximum.

In the first case, it can be assumed that the entire scanning field 12 has a tone value of 128, ie each sampling point 14 of the scanning field 12 has a tone value of 128. The recorded quasi-halftone image must therefore have 128 colored recording points in the area corresponding to the scanning field. For all of the smallest subfields Fb _{i, j} (see FIG. 1b), a tone value sum of

The carry bit of the tonal value sum is "L", ie one recording point is to be printed per smallest subfield Fb _{i, j} . After the first stage of the method, the recording dots shown in Fig. 2b are printed.

According to the above statements, the sum of the tonal value sums of the two smallest subfields combined to form a second order field is determined for the tonal value sums of the fields Fc _{i, j of the} second order ( FIG. 1c), the carry bit not taking into account the tone value sums of the smallest subfields remains. This results in second order for the tonal value sums of the fields Fc _{i, j}

Since the carry bit takes the value "0", the Second order fields no record point printed.

Exactly the same considerations allow for the fields third ( Fig. 1b), fourth ( Fig. 1e), fifth ( Fig. 1f), sixth ( Fig. 1g), seventh ( Fig. 1h) and eighth ( Fig. 1i) order hire. In all cases, the carry bit of the tone value has the value "0", which means that none of the recording dots belonging to the individual higher order subfields is printed. A total of 128 recording points are printed, namely exactly one recording point in each of the 128 smallest subfields, corresponding to the tonal value sum for each smallest subfield. The critical tone value of 128 for the entire scanning field 12 can thus be reproduced exactly in the recorded image.

Now it is assumed that each sample point 14 of the sample field 12 has the tone value 127. The following then applies to the tone value sum of each smallest subfield Fd _{i, j} :

The carry bit of the tonal value of each smallest subfield is "0". Therefore, none of the a recording point is printed in the smallest subfields.

For the tone value sum of the subfields Fc _{i, j of the} second order ( FIG. 1c), the following results:

The carry bit of the tone value sum of each second-order subfield thus has the value "L", so that a recording dot is printed in each second-order subfield in accordance with the pattern shown in FIG. 2c. According to the second stage of the process, 64 recording dots are already printed.

For the tonal value sums of the higher-order subfields (FIGS . 1d to 1i), the following results:

(32 further recording dots are printed)

(16 further recording dots are printed)  

(8 more recording dots are printed)

(4 more recording points are printed)

(2 more recording dots are printed)

(1 further recording point is printed)

Altogether, after the second stage 64, after the third stage 32, after the fourth stage 16, after the fifth stage 8, after the sixth stage 4 and after the seventh stage 2, further recording dots are printed, with a further one after the eighth stage Recording point is printed. In total, 127 of the 256 recording points corresponding to the scanning field 12 are printed. In this case too, the tone value 127 of the scanning field is reproduced exactly.

The two examples discussed show how exactly that Process works without considering errors of any kind are required.

The examples discussed so far have always been cases in which the recording fineness was equal to the scanning fineness, that is, a sampling point was represented by a recording point. In FIG. 4, three examples are given in which the recording definition is times greater than the scan definition, which means that a sample point is represented by two recording points (Fig. 4a to 4d). In Figures 4e. 4f and an example is shown in which the recording definition twice as great as the scan definition, which means that a sample point is represented by four recording points.

In the in Figs. 4a and 4b illustrated case, each of the 128 sampling points of the scan field 14 is a 12 a through the two to a smallest subfield Fb _{i, j} belonging recording dots 16 reproduced. The mean tone value for a smallest subfield is therefore equal to the tone value of the sampling point 14 a filling this smallest subfield. With such a position of the scanning point in relation to the smallest subfields, the smallest subfields need not be summed up; the carry bit, on the basis of which a recording point belonging to a smallest subfield is printed, corresponds to the most significant bit of the binary-coded tone value of the relevant sampling point. In the first process stage, 128 addition steps, ie approximately 50% of the addition steps required for the entire scanning field, can be saved. With regard to the further method steps, the procedure can be analogous to the example described in FIG. 3. The tone value sum (not calculated) required for the further calculations for the smallest subfields is generated from the binary-coded tone value of the corresponding sampling point by adding a "0" as the least significant bit. The most significant bit of this nine-digit binary code corresponds to the carry bit of the tone value for the smallest subfields and is not taken into account for the subsequent calculations.

Recording images with a recording to sample fineness ratio of has the advantage that less data is generated on the scanning side, the scanning but is still sufficiently sharp. The opposite of the Scanning higher recording fineness causes the Image has less granularity, which makes it appears calmer and is easier to look at. In relation the processing of the tonal value data results in the Advantage that 50% fewer adding steps per scanning field what is required in relation to the apparatus Realization corresponds to a saving of 128 adders.

An example is also shown in FIGS. 4c and 4d, in which the recording fineness is greater by a factor than the scanning fineness. Here, however, the scanning points 14 c of the scanning field 12 c are rotated by 90 ° with respect to the smallest subfields Fb _{i, j} . A sampling point 14 c of the scan field 12 is no longer represented by two to one and the same smallest subfield belonging recording dots 16, but two by two adjacent recording points 16 adjacent the smallest subfields. The method can proceed exactly as described in connection with FIG. 3 if the left half of a sampling point 14 c is considered to belong to a smallest subfield and the right half to belong to the adjacent right smallest subfield.

In the case shown in FIGS. 4e and 4f, the recording fineness is twice the scanning fineness. A scanning point 14 e of the scanning field 12 e is represented by the four recording points 16 belonging to two smallest subfields. In this constellation, the totals for the smallest subfields Fb _{i, j} (see Fig. 4f and Fig. 1b) and for the subfields Fc _{i, j of the} second order (see Fig. 1c) can be replaced by using the two most significant bits of the tone value of the sampling point 14 e considered. The most significant bit corresponds to the carry bit for the otherwise necessary dot formation for the smallest subfields, while the second most significant bit corresponds to the carry bit for dot formation for the second order subfields. The further stages of the method run in the manner described in connection with FIG. 3.

FIG. 5 shows the block diagram of the device for executing the previously described method. A (not shown) sensor device, such as. B. a scanner, scans the template and outputs an analog image signal for each sampling point, which is digitized in an analog / digital converter, also not shown, and is entered via the data flow path 22 into a data input memory 24 . The scanner outputs control signals to a central control unit (microprocessor) 28 via an image control path 266 . These control signals are e.g. B. the beginning of the picture, end of picture and end of line signal. The microprocessor 28 controls the data input memory 24 via a control line 30 in such a way that the data received via the data flow path 22 are stored in individual data blocks. All the tone value data of a scanning field, e.g. B. 256 8-digit binary code, stored. The control signals output by the microprocessor as a function of the signals on the image control path 26 are generated in accordance with a control program stored in a ROM memory 30 .

When all the tone value data belonging to a scanning field are stored as a block in the data input memory 24 , the data input memory passes on all 256 tone value data signals to an input switching network 34 connected to the data input memory 24 via the output lines 32 upon a signal from the microprocessor 28 . The tone value data signals present on the lines 32 as input signals of the input switching network are distributed over the output lines 36 of the input switching network 34 in such a way that the tone value data belonging to the smallest subfields are present on adjacent output lines of the input switching network. The input switching network thus connects its inputs to its outputs in such a way that the tone value data belonging to a smallest subfield are present at adjacent outputs, the tone value data signals belonging to a second order subfield are present at adjacent pairs of outputs, etc. The input switching network essentially consists of a network-controlled switch with which, depending on the control, the inputs and the outputs of the input switching network can be connected to one another in any way. The individual switches are activated by the microprocessor 28 , which is connected to the input switching network 34 via control lines 38 . A plurality of control patterns for the input switching network 34 are stored in the ROM 30 , so that the control of the switches and thus the combination of the tone value data into the individual data blocks of different sizes can be changed from scanning field to scanning field. The division of the tone value data into the individual blocks of different sizes corresponds to the division of the scanning field into the sub-fields of different sizes.

The "shuffled" in the input switching network 34 tonal value data are passed in parallel to an arithmetic processor 40, which is connected to the output lines 36 of the Eingangsumschaltnetzwerks 34th The arithmetic processor 40 , which consists of a cascade-shaped network of individual interconnected digital adders, will be described in more detail later with reference to FIG. 6. The arithmetic processor 40 carries out the addition steps and outputs a binary output signal per data block, which indicates whether or not a recording point is to be printed in a subfield corresponding to a data block. The signals output for each data block are passed on via output lines 42 as input signals to an output switching network 44 connected to the output lines 42 .

The output switching network 44 , like the input switching network 34, consists of a network of interconnected controlled switches which connect the inputs of the output switching network with its outputs in accordance with the actuation of the switches. The control signals for the output switching network 44 , ie for the controlled switches of the output switching network, are generated in accordance with the control patterns stored in the ROM 30 by the microprocessor 28 and passed on to the output switching network 44 via the control lines 38 , via which the input switching network 34 is also controlled. The number of outputs of the output switching network 44 is equal to the number of recording points per scanning field. The distribution shown in Fig. 2, it sets, for example, the function of the respective mean values to be printed recording points based on, then the switch matrix of the Ausgangsumschaltnetzwerks 44 is driven such that each input line for a smallest data block (for a smallest subfield Fb _{i j,} of Figure . 1b) respectively connected to the output of Ausgangsumschaltnetzwerks 44 for the inside of the smallest sub-field to be printed recording dot. In the same way, the inputs for the higher-level data blocks are connected to the corresponding outputs.

The binary output signals of the output switching network 44 are fed to a data output memory 48 via output lines 46 . The data output memory 48 is controlled by the microprocessor 28 ; the control signals required for this are sent from the microprocessor 28 to the data output memory 48 via control lines 50 . In the data output memory 48 controlled by the microprocessor 28 , the binary signals arriving per scanning field via the output lines 46 are stored in accordance with the division of the front position into the scanning fields. If all the binary data for an original are available, they are forwarded to the recording device (not shown) via the output line 52 of the data output memory 48 in response to a signal from the microprocessor 28 .

For the following considerations, assume that a scan field consists of 16 scan points and is represented by 16 record points. The (analog) tone value of a sampling point is then converted into a four-digit digital signal in the analog / digital converter. This means that 16 tone values (from 0 to 15) can be binary coded. The data input memory 24 receives four-digit digital signals per scanning field 16 , which it forwards via its output lines 32 to the input switching network 34 . With parallel data processing, the input switching network has 16 × 4 inputs and 16 × 4 outputs. As shown in FIG. 7, the scanning field consisting of the 16 sampling points is broken down into a total of 15 subfields, ie the arithmetic processor has 15 outputs, namely one for each subfield (each data block). The arithmetic processor 40 thus generates 15 binary output signals from the 16 × 4 binary input signals. In the output switching network, 16 binary output signals are generated from these 15 binary output signals of the arithmetic processor, namely an output signal for each of the 16 recording points corresponding to the scanning field. Accordingly, the data output memory 48 requires 16 individual memory cells per scanning field, whereas the data input memory 24 requires four times per scanning field. The amount of data has therefore decreased by a factor of 4. More generally, the amount of data is reduced by the square root of the sampling or recording points combined into a scanning field.

The structure of the arithmetic processor is described with reference to FIG. 6, specifically in the event that 16 recording points each belong to a scanning field. However, 15 recording points are sufficient to display the 16 tonal values. The recording point designated by reference number 19 in FIG. 7f is not required. However, this does not have a disruptive effect for the reason given in connection with FIG. 2 for the reproduction. The representations in FIGS. 7 and 8 are greatly enlarged; in fact, the recording points each have a diameter of 20 µm.

The scanning field shown in FIG. 7a and consisting of the scanning points P 1 to P 16 is, as shown in FIGS. 7b to 7f, subdivided into a total of 15 partial fields which are identified by the reference symbols drawn in the figures.

The arithmetic processor 40 consists of a network of adders 54 connected in cascade. The first “cascade stage” b has as many adders 54 as there are smallest subfields or data blocks. In the present example, eight smallest data blocks F 17 to F 24 are available. In the adders of the first stage b , the tone value sum from the tone values of the two tone values belonging to a smallest data block is added. The second stage b of the cascade has as many adders 54 as there are subfields (data blocks) of the second order. According to FIG. 7c, there are four. The inputs of the adders 54 of the second stage c are connected to the outputs of those adders 54 of the first stage b which calculate the tonal value sums of the two smallest data blocks combined to form the second-order data block. The third stage d of the cascade includes only two adders 54 which, according to the combination of second-order data blocks ( FIG. 7c) and third-order data blocks ( FIG. 7d) shown in FIGS. 7c and 7d, with the outputs of adders 54 are connected to level c . In the last cascade stage e there is only one adder 54 , the inputs of which are connected to the outputs of the two adders 54 of stage d . The adders 54 shown in FIG. 6 are 4-bit digital adders which have 2 × 4 inputs 56 , four outputs 58 and a carry output 60 . The connecting lines of the adders 54 shown by thick lines in FIG. 6 each set up four parallel lines, while the thinly drawn lines are individual connections.

The 16 4-bit lines connected to the adders 54 of the first cascade stage b represent the inputs 36 of the arithmetic processor 40. The carry outputs 60 of the adders 54 correspond to the outputs 42 of the arithmetic processor 40 , with each carry output 60 is assigned to a data block (subfield). FIG. 6 shows which output belongs to which data block identified in FIG. 7. So that, for example, the carry output 60 of the top adder 54 of the first cascade stage b corresponds to the output for the block F 24 , the tonal value data for the sampling points P 12 and P 16 must be present at the inputs 56 of the adder 54 ( FIG. 7). The tone value data at which inputs 36 of the Ar 07872 00070 552 001000280000000200012000285910776100040 0002003718521 00004 07753ithmetic processor 40 must be present is shown in FIG. 6. As a result of the activation by the microprocessor 28 , the input switching network 34 causes the tone value data signals to be re-sorted in a manner corresponding to the division of the scanning field into subfields and to be grouped together.

The operation of the arithmetic processor 40 is explained using an example. In Fig. 8a of the corresponding tonal value is specified for each of the 16 sample points. Fig. 8b shows the distribution of the recording dots to be printed when the scanning field is reproduced with the tonal values indicated in Fig. 8a. The calculation given below is analogous to the calculation carried out in connection with FIG. 3.

In evaluation of the respective resulting from the summation of the carry bits, the distribution indicated in Fig. 7, the distribution of the printed within the scanning to the recording dots shown in Fig. 8b results, taking into account each sub-field to be printed recording dots.

FIG. 9 shows which digital signals are present on the individual lines of the arithmetic processor 40 for the example shown in FIG. 8a. As can be seen, a binary signal is present on the individual output lines 42 of the arithmetic processor, which corresponds to the carry bit of the tone value sum of the associated data block. Although in the cascade stages c, d and e each do not process the complete tonal value sums, but only the last four bits of the tonal value sums, this leads to the correct output data. The microprocessor 28 drives the output switching network 44 in such a way that the input of the output switching network 44 , at which the output signal for a data block is present, is connected to the output of the output switching network 44 for the recording point which is to be printed in dependence on the amount of the tone value of the data block is.

FIGS. 5 and 6 show the circuit realization of the method described above. The embodiment shown in the two figures with four cascade levels is also of very special interest if the quality requirements for the reproduction are somewhat lower. The scanning-side image signals from scanners in the office communication area, e.g. B. from digital copiers or Telefaksimile devices have a much lower signal / noise ratio than is the case with the devices for the graphics industry. If the template is divided into scanning fields with 16 × 16 = 256 scanning points each, the eight-stage cascade connection of the arithmetic processor 40 required for processing would practically only evaluate noise signals on the last cascade stages, which is uneconomical. Instead, it makes sense to optimize the number of cascade levels to the number of "noise-free image signal bits". Four cascade levels are particularly economical and qualitatively acceptable for a number of requirements.

The special economy results from that then 4-bit adders are sufficient to the Perform calculations according to the procedure described above, and that very few additions, namely 1 + 2 + 4 + 8 = 15 are enough to the four-stage Complete cascade. Persist then both the scanning-side matrix of scanning signals, which also only contain 4 bits each, as also the recording point matrix on the recording side only of 16 elements, so that only very small memory necessary or larger buffer storage simply too are to be realized and managed.

As a result of the high economic efficiency and the low technical outlay, it is particularly advantageous to directly interconnect the total of 15 adders without a buffer, as shown in FIG. 6. As a result of the direct interconnection or integration into a module, the calculations for a matrix can be reliably guaranteed in significantly less than 100 ns.

. The complex adder of FIG 6 can also replace the four upper cascade stages of an eight-stage cascade in an advantageous manner, because there it is sufficient, the noise components - to separate after the fourth cascade stage and only the actually evaluated 4 bits - given by the 4 low-order bits higher order to process further.

Each scanning field can use the same division into subfields and the same distribution of recording dots to be printed per subfield, if the demands on the quality of the reproduction are not too great. In such a case, the input and output switching network can be implemented by hard wiring their inputs and outputs. When using one and the same subdivision of the scanning field and distribution of the recording dots to be printed, however, patterns arise in the reproduction, which are undesirable in particular in the colored printout. The patterns of the individual process colors overlap and form interference patterns, so-called moir ', whose structures are significantly larger than those of the individual colors. It is therefore better if the division of the scanning field and the distribution of the recording points to be printed (the addressing of the scanning and the recording points) are changed from scanning field to scanning field, thereby avoiding the occurrence of undesired structures. The respective types of addressing can be selected according to a random or quasi-random principle. The various types of addressing are stored in the device 30 shown in FIGS. 5 and 6 in the ROM 30 which is accessed by the microprocessor 28 . Both for the subdivision of the scanning field (addressing on the input side), for which an example is given in FIG. 1, and for the distribution of the recording points to be printed per subfield (addressing on the output side), for which an example is given in FIG. 2 there are several solutions that take different practical requirements into account to different degrees. In the device according to FIG. 5, the different addressings can be implemented with the aid of the two networks 34 and 44 controlled by the microprocessor 28 .

The previous description of the method and the device relates to the case where the original is recorded with the same brightness. However, if it is to be reproduced darker or lighter and / or with higher contrast or lower contrast, it is advantageous to first convert the (analog) tone value signals on the scanning side before further processing in accordance with a corresponding assignment table or characteristic curve. Then, namely, the recording process can again be carried out with the gray tone 128 as a reference value and the data conversion device according to FIGS . 5 and 6 can be used accordingly. All of the simplifications previously derived and specified thus also apply to such manipulated recording images.

Claims (30)

1. A method for recording quasi-halftone images, in which an original is reproduced on a recording medium by individual recording elements ( 16 ) with the color value "full" or "zero", the original divided into a plurality of individual scanning elements ( 14 ) consisting of scanning fields ( 12 ) and each scanning field ( 12 ) is successively subdivided into the smallest subfields (Fb) with a predetermined size, characterized in that for each smallest subfield (Fb) and for each of the superordinate subfields to which the smallest subfield in question belongs, from the tonal values the mean value of the scanning elements ( 14 ) located in the subfield under consideration is determined or estimated and that at least one of the recording elements ( 16 ) associated with the subfield under consideration is printed if the mean value is greater than a predetermined reference tone value. 2. The method according to claim 1, characterized in that the position of the recording elements ( 16 ) to be printed as a function of the amounts of the mean values is determined within the subfields of the same order in a random selection. 3. The method according to claim 1 or 2, characterized in that the position of the recording elements ( 16 ) to be printed as a function of the amounts of the mean values in the subfields of different orders is determined such that each recording element ( 16 ) is printed once. 4. The method according to one or more of claims 1 to 3, characterized in that for copying the original with a tonal value gradation which is coarser than that of the original, the recording elements ( 16 ) to be printed as a function of the mean tonal values of the scanning elements ( 14 ) ) are only printed for those subfields whose number of scanning or recording elements ( 14 or 16 ) is greater than or equal to the number required to display the original with the desired tonal value gradation. 5. The method according to one or more of claims 1 to 4, characterized in that each scanning field ( 12 ) is successively divided into sub-fields of equal size and that the number of sub-fields into which each is subdivided is the same. 6. The method according to one or more of claims 1 to 5, characterized in that the arrangement of the subfields within the scanning field ( 12 ) is regular. 7. The method according to one or more of claims 1 to 6, characterized in that 4 ⁿ 4 scanning elements ( 14 ) are combined in a 2 ⁿ rows and 2 ⁿ columns scanning field ( 12 ) and that each is divided into two sub-fields of lower order. 8. The method according to one or more of claims 1 to 6, characterized in that 4 ⁿ 4 scanning elements ( 14 ) in a 2 ⁿ rows and 2 ⁿ columns having scanning field ( 12 ) are combined and that each is divided into four sub-fields of lower order . 9. The method according to one or more of the claims 1 to 8, characterized in that the size the smallest subfields from the ratio Corresponding recording to scanning fineness. 10. The method according to one or more of claims 1 to 8, characterized in that the recording fineness is chosen greater than the scanning fineness and that the size of the smallest subfields (Fb) is equal to the size of two recording elements ( 16 ). 11. The method according to claim 10, characterized in that for those subfields that are smaller than or as large as a scanning element ( 14 ), the tonal value of the scanning element ( 14 ) in which the subfield is located is taken as an average. 12. The method according to one or more of claims 1 to 10, characterized in that the scanning and the recording elements ( 14 and 16 ) are each arranged in an orthogonal matrix. 13. The method according to one or more of claims 1 to 11, characterized in that the scanning and the recording elements ( 14 and 16 ) are each arranged hexagonally. 14. Method for recording multicolored quasi-halftone images, which consist of several additives or subtractive basic colors are put together, characterized, that for each primary color the procedure according to one of claims 1 to 13 is carried out and the the resulting single-color quasi-halftone images be overlaid. 15. Method of recording three-dimensional Quasi halftone images, with a template on a recording medium by individual recording volume elements with the color value "full" or "zero" is rendered three-dimensionally, the template into several of individual sample volume elements existing sample volumes divided and each sample volume successively into smallest partial volumes with a given Volume is divided characterized, that for every smallest partial volume as well as for each of the parent sub-volumes to which the smallest sub-volume in question belongs from the tonal values of those in the sub-volume under consideration lying sample volume elements the mean is determined or estimated and that in each case at least one of the recording volume elements, assigned to the sub-volume under consideration are reproduced with the color value "full"  if the mean is greater than a predetermined reference value. 16. Device for converting dot-tone values, with a processing unit which converts the tonal values of a template divided into a plurality of scanning fields ( 12 ) into recording pixel data which correspond to either the color value "full" or "zero", characterized in that the processing unit :
an input switching network ( 34 ) which receives the pixel-tone value data and combines them into sub-fields according to pixel-tone value data blocks by dividing the scanning fields ( 12 ) of the original,
an arithmetic processor ( 40 ) connected to the input switching network ( 34 ), which outputs an output signal for each block when the average tone value of the block is greater than a predetermined reference value, and
an output switching network ( 44 ) connected to the output of the arithmetic processor ( 40 ), which, depending on the output signals of the arithmetic processor ( 40 ), records the recording pixel data or the distributions of the recording pixel data corresponding to the color value "full" and the color value "zero" created. 17. The apparatus according to claim 16, characterized in that the arithmetic processor ( 40 ) has a multi-level network of interconnected adders ( 54 ) with carry output ( 60 ), in the lowest stage (b) as many adders ( 54 ) as the smallest data blocks are provided and each adder ( 54 ) of the lowest level (b) adds the tonal value data of the pixels belonging to a smallest data block, and that the outputs ( 58 ) of the adders ( 54 ) of a level with the inputs ( 56 ) of the adders ( 54 ) are connected to the next higher level and the carry outputs ( 60 ) of the adders ( 54 ) of each level are the outputs of the arithmetic processor ( 40 ). 18. The apparatus according to claim 16 or 17, characterized in that a control unit ( 28 ) is provided which controls the input switching network ( 34 ) such that the inputs for the pixel tone value data are connected to the outputs in accordance with the division into data blocks, and that the control unit ( 28 ) controls the output switching network ( 44 ) such that the inputs are each connected to those outputs at which the data for the recording pixels with the color value "full" are present. 19. The device according to one or more of claims 16 to 18, characterized in that the type of combination of the pixel tone data into data blocks and the type of distribution of the recording pixels are stored in a memory ( 30 ) connected to the control unit ( 28 ). 20. The apparatus according to claim 19, characterized in that in the memory ( 30 ) a plurality of different summarization and distribution patterns are stored. 21. The device according to one or more of claims 16 to 20, characterized in that a data input memory ( 24 ) controlled by the control unit ( 28 ), in which the pixel tone value data of the template are temporarily stored and which supplies the tone switching data to the input switching network scanning field by scanning field , and a data output memory ( 48 ) controlled by the control unit ( 28 ) are provided, in which the recording pixel data output by the output switching network ( 44 ) per scanning field are stored. 22. The device according to one or more of claims 16 to 21, characterized in that at the inputs of the input switching network ( 34 ) in each case the pixel tone value data of a scanning field ( 12 ), the pixels arranged in a matrix with 2 ⁿ rows and 2 ⁿ columns , n is a natural number and the tonal values of the pixels are present as a binary number with 2 n digits, and that the adders ( 54 ) of the arithmetic processor ( 40 ) are digital adders for adding 2 n -digit binary numbers. 23. The apparatus according to claim 22, characterized in that the number of carry outputs ( 60 ) of each adder ( 54 ) corresponds to the number of pixel dot data per smallest data block or the number of subordinate data blocks belonging to a higher-level data block. 24. The apparatus of claim 22 or 23, characterized in that when the recording fineness is greater than the scanning fineness, the input lines of the arithmetic processor ( 40 ) for the pixel tone data with the inputs ( 56 ) of the adders ( 54 ) of higher levels can be connected, the input lines of the arithmetic processor ( 40 ) for the most significant bits of the pixel tone data being additionally connectable to the outputs of the arithmetic processor ( 40 ). 25. The apparatus according to claim 24, characterized in that with a-times greater recording fineness as a scanning unit, the input lines of the arithmetic processor ( 40 ) for the pixel tone data with the inputs ( 56 ) of the adders ( 54 ) of the second stage (c) and in addition the input line of the arithmetic processor ( 40 ) for the most significant bit of the pixel tone value data can be connected to the outputs of the arithmetic processor ( 40 ). 26. The apparatus according to claim 24, characterized in that with a 2 times greater recording fineness as the scanning fineness, the input lines of the arithmetic processor ( 40 ) for the pixel tone data with the inputs ( 56 ) of the adders ( 54 ) of the third stage (c) and the input lines of the arithmetic processor ( 40 ) for the most significant bits with the outputs of the arithmetic processor ( 40 ) for the smallest data blocks and the input lines of the arithmetic processor ( 40 ) for the second most significant bits with the outputs of the arithmetic processor ( 40 ) for which data blocks superordinate to the smallest data blocks can be connected. 27. The device according to one or more of the claims 16 to 23, characterized in that, if the recording fineness is greater than or is the same size as the scanning fineness that smallest data blocks each two tone value data include. 28. The apparatus according to claim 27, characterized in that the control unit ( 28 ) controls the input switching network ( 34 ) such that each input for a tone data signal is connected to a number of outputs corresponding to the ratio of recording to scanning fineness. 29. The device according to claim 28, characterized in that at a ratio of one Input for a tone value data signal with two outputs connected is. 30. The device according to claim 28, characterized in that at a ratio of 2 a Input for a tone value data signal with four outputs connected is.